DOP79 Biomarkers for IBD using OLINK Proteomics inflammation panel: Preliminary results from the COLLIBRI consortium

Abstract

i123 Conclusion: In clinical practice, a relatively high proportion of Crohn's disease patients who received ustekinumab or vedolizumab for anti-TNF failure, maintained these drugs in the medium-long term, although ustekinumab retention rate was higher in comparison with vedolizumab. Background: Circulating serum proteins have provided insights into disease pathogenesis and are being used to identify prognostic, diagnostic and therapeutic biomarkers for chronic inflammatory diseases. With this pilot project, the Collaborative IBD Biomarker Research Initiative (COLLIBRI) consortium aimed to unravel disease heterogen-eity in inflammatory bowel disease (IBD). Methods: Serum samples were cross-sectionally obtained from 3,390 individuals (Crohn's disease (CD), n=1815; ulcerative colitis (UC), n=1170; healthy, n=405) recruited at nine centres from Sweden and Belgium. Relative levels of 92 proteins were analysed using the Proseek Multiplex Inflammation I Probe kit 96x96 (Olink Proteomics, Uppsala, Sweden) and reported as arbitrary units, i.e., normalised protein expression on a log2 scale. Using a multivariate integrative approach, we identified protein signatures distinguishing CD and UC samples and attempted to identify clusters or subgroups within patients. Recruiting centre, cohort and batch information were considered for the integrative analysis. Optimisation was performed for identifying the number of components and features per component using 5-fold cross-validation and Leave-One-Group-Out-Cross-Validation, respectively. Information on transcriptional regulators was retrieved from the ReMap project using the orthogonal regulatory resource ChEA3. Results: A panel of 8 proteins was identified which could segregate CD and UC patients (Figure 1). FGF19 exhibited a consistent trend of expression (downregulated in CD) across all batches of datasets. An integrated AUC of 72.5% was achieved across the different batches of samples used in the study with the highest AUC (79.2%, P-value 8.5e-07) being recorded for a single batch of samples (CD = 42, UC = 56). On a centre-specific dataset, the cross-centre integrated signature achieved an AUC of 75.1%. We identified three transcription factors (MEF2A, BATF, NFKB2), of which the two latter ones are known to modulate intestinal inflammation and which could potentially regulate